摘要
应用计算机视觉技术对温室白粉虱自动计数技术进行了研究。采用胶卷照相机和家用摄像机对田间温室白粉虱寄生的叶片进行拍摄 ,以获得其数字图象。对白粉虱图象的分割采用 Johannsen基于熵的分割算法。对分割后的二值图象利用区域标记算法得到白粉虱个体的数量。对叶片上挨在一起的白粉虱个体采用数学形态学算法进行了分离。用 1 9个带虫叶片样本的统计结果表明 ,直接利用分割图象进行白粉虱个体计数的累积准确率达 91 .99% ,而分离处理的算法则需要改进。因此 ,这一技术具有进一步在生态研究和 IPM实践中推广的可能性 ,这将使田间微小昆虫的种群数量监测和调查的工作量大幅度降低 ,而准确率得到显著提高。
The automated counting technology was developed for the tiny insects using computer vision technology and aiming to overcome counting difficulty when sampling the insects in the field The objective insect selected was Trialeurodes vaporariorum (Westwood) The digital image of the leaf on which T vaporariorum fed was acquired using a VHS CCD camera and a frame grabber The image was segmented employing Johannsen's entropy segmentation algorithm A region labeling algorithm was applied to the segmented binary image to count the insect number Sometimes a few insects were connected to each other,so a disconnect algorithm was developed based on the morphological operations An accumulative accuracy of 91 99% was achieved when the automated counting method was tested on 19 leaf images, and it demonstrated a promising potentiality of this method for ecological and IPM applications.It can save labor and time and enhance the quality of pest monitoring
出处
《生态学报》
CAS
CSCD
北大核心
2001年第1期94-99,共6页
Acta Ecologica Sinica
基金
国家自然科学基金!( 3 984 0 0 0 4 )
国家高技术研究发展计划课题!( 863 -3 0 6-ZD0 5-0 2 -0 3 )
高等学校博士点专项科研基金联
关键词
自动计数
图像处理
计算机视觉
温室白粉虱
害虫
automated counting
image processing
computer vision
trialeurodes vaporariorum (Westwood)H
